February 2018 – Auto-Segmentation of Claims

By Jonathan Polon

The application of predictive modeling to segment disability claims has come to be considered best practice since the approach was first introduced in the early 2000s. This presents an opportunity to improve current claims management practices for Short-Term Disability claims by developing an auto-segmentation model.

JP4

As Short-Term Disability claims are very transactional in nature – characterized by high-volumes and short-durations – the vast majority of claimants will return to work with minimal or no active case management.  An auto-segmentation model allows disability case managers to focus their resources on the highest-leverage claims.

Low Complexity Claims

Most STD claims are low complexity and will return to work with minimal intervention. Claims in this segment can be pre-approved for 2-8 week of benefits. The pre-approval period is unique to each claim and depends upon each claim’s specific attributes and predicted duration. The pre-approval period serves to define an expected date for return to work for both the employee and the employer to strive toward. It also provides an opportunity for any required accommodations to be identified and put in place.

Not all low complexity claims will resolve during the pre-approval period. For these cases, the case manager will need to communicate with both the employee and employer to assess the situation. In many cases, the claimant will simply need an extra couple of weeks beyond the pre-approval period to return-to-work. In other cases, the claimant may require active case management to try and facilitate return-to-work prior to the end of the STD benefit period.

High Complexity Claims

There are some STD claims that can be identified at the time of intake as being very unlikely to resolve during the STD benefit period. These claims can receive a case management assessment immediately upon intake so that a case management plan can be designed and implemented. Oftentimes, the insurer may even decide to refer the claimant to Ling-Term Disability case management – even during the STD benefit period.

Data Sources

Structured Data

The auto-segmentation models are predictive models trained with an insurer’s historic claim data. The training process uses the data fields that are readily available from the insurer’s claim system. For example: age, gender, diagnosis, occupation, etc.

Free-form Text Data

The models can be enhanced by analyzing free-form text data, if available. Examples of free-form text data includes: claim applications, attending physician statements and case management notes.

Claimant Questionnaires

The models can also be enhanced by using questionnaires as a means to learn more about each claimant, their health and their mindset. Questionnaires are particularly useful when applied to complex claims. This is where a deeper understanding of the claimant’s condition, mental health, or motivation to return to work would significantly assist the model’s ability to differentiate these claims and provide insight in the way that the claim is managed. This is an area where The Claim Lab has specific expertise.

Summary

Auto-segmentation models are great tools to enhance the productivity of Short-Term Disability case management teams. The models allow the case managers to focus their efforts on the claims where active intervention can have the greatest impact in improving the likelihood of return-to-work.

For more information, please email us at  info@claimlab.org